Experiment tracking system

One or more non-transitory computer-readable media storing instructions executable by one or more processors that, when executed, cause the one or more processors to perform acts comprising:

• receiving, by a machine learning (ML) experiment system, an experiment request associated with an ML model including a first hyperparameter and a second hyperparameter, wherein the first hyperparameter is an experiment hyperparameter, the second hyperparameter has a static value, and the experiment request includes a plurality of values of the first hyperparameter and the static value of the second hyperparameter;
• generating, by the ML experiment system, a configurable hyperparameter set of the ML model including the first hyperparameter and the second hyperparameter;
• training, by the ML experiment system, the ML model with the first hyperparameter having a value of the plurality of values of the first hyperparameter and the second hyperparameter having the static value to generate an experimental result corresponding to a version of the configurable hyperparameter set with the first hyperparameter of the configurable hyperparameter set having the value and the second hyperparameter of the configurable hyperparameter set having the static value, the experimental result including a metric value of a metric determined based on the training of the ML model with the first hyperparameter having the value of the plurality of values of the first hyperparameter and the second hyperparameter having the static value;
• and
• storing a queryable association of the configurable hyperparameter set and the experimental result in an experiment tracking datastore such that the experimental result is retrievable based on a query including one or more query parameters that match values of one or more corresponding configurable hyperparameters of the version of the configurable hyperparameter set.

A system comprising:

• one or more processors;
• and
• one or more non-transitory computer-readable media storing instructions executable by the one or more processors, wherein the instructions, when executed, cause the one or more processors to perform acts comprising:
• receiving, by a machine learning (ML) experiment system, an experiment request associated with an ML model including a first hyperparameter and a second hyperparameter, wherein the first hyperparameter is an experiment hyperparameter, the second hyperparameter has a static value, and the experiment request includes a plurality of values of the first hyperparameter and the static value of the second hyperparameter;
• generating, by the ML experiment system, a configurable hyperparameter set of the ML model including the first hyperparameter and the second hyperparameter;
• training, by the ML experiment system, the ML model with the first hyperparameter having a value of the plurality of values of the first hyperparameter and the second hyperparameter having the static value to generate an experimental result corresponding to a version of the configurable hyperparameter set with the first hyperparameter of the configurable hyperparameter set having the value and the second hyperparameter of the configurable hyperparameter set having the static value;
• and
• storing a queryable association of the configurable hyperparameter set and the experimental result in an experiment tracking datastore such that the experimental result is retrievable based on a query including one or more query parameters that match values of one or more corresponding configurable hyperparameters of the version of the configurable hyperparameter.